Skip to content
ACN Claude Toolkit

Roadmap

V1 is in Preview — shipped, in pilot use, behavior may change based on feedback before GA. This page lists what’s done, what’s queued, and what’s deliberately out of scope.

V1 — Preview (shipped)

Section titled “V1 — Preview (shipped)”
  • acn-base-sdlc — 4 skills, 5 rules, 3 agents
  • acn-pursuit-toolkit — 2 skills, 2 rules, 1 agent, 3 templates
  • acn-slide-gen — 1 skill, 2 flavors, 3 helper scripts

Distribution: --plugin-dir from a local clone (the CIO allowlist doesn’t yet permit /plugin marketplace add against the CodeCommit URL).

V1.5 — queued

Section titled “V1.5 — queued”

These are decided in scope and waiting on time:

  • More SDLC skills — adr, dockerfile, runbook, integration-test — migrated from the prior POC into acn-base-sdlc
  • Real CI validators — validate-marketplace.ts, validate-frontmatter.ts, check-marketplace-sorted.ts — adapted from anthropics/claude-plugins-official. See CI Validators.
  • Branch protection on main — the approval-rule template at approval-rules/main-branch.json gets attached. V1 leaves this off so the solo maintainer can self-ship.

V2 — under consideration

Section titled “V2 — under consideration”

These depend on demand or external unblockers:

  • Marketplace add path — replaces --plugin-dir once the CIO allowlists the toolkit’s CodeCommit URL. The directory layout already matches Anthropic’s claude-plugins-official reference, so no plugin-side changes needed; the install command becomes /plugin marketplace add codecommit::us-east-2://magellan@acn-claude-toolkit.
  • Vertical overlays — acn-fsi-compliance, acn-hls-compliance, acn-pubsec-compliance for regulated-industry workflows. Each as a separate plugin layered on top of acn-base-sdlc. Decision gates on demand from active engagements.

What V1 deliberately doesn’t include

Section titled “What V1 deliberately doesn’t include”

Some additions are technically possible but out of scope by design:

  • Format-on-save hooks — conflicts with team Prettier/ESLint configs
  • Auto-fix skills that skip human review — every code-modifying skill in V1 surfaces changes to the user before applying
  • Client-specific assets (templates, brand kits, sample artifacts) — stays on the FDE workstation as user-scope content, not in the marketplace
  • FDE-personal context (meeting notes, prior conversations, research) — lives in ~/.claude/skills/ user scope

See Contributing for the full list of “what NOT to add.”

Feedback drives the roadmap

Section titled “Feedback drives the roadmap”

V1 is a pilot. The V1.5 / V2 mix tilts based on what pilot users actually ask for. Channels:

  • Immediate friction — message Felix on Teams or Slack
  • Feature requests / bugs — CodeCommit issue
  • New skill proposals — follow Adding a Skill; the litmus test is whether wrapping the guidance in a skill changes Claude’s output